Purchase this article with an account.
Daniel Wollschläger, Barton L. Anderson; Scission causes large color-induction effects in textured center-surround stimuli. Journal of Vision 2007;7(9):789. doi: 10.1167/7.9.789.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Our ability to perceive invariant object colors requires mechanisms that disentangle object properties from context dependent influences like illumination, inter-reflections, and transparency. Previous work has shown that varying the mean and the covariance structure of surround colors induces changes in the appearance of embedded targets. Such phenomena are usually explained with low-level mechanisms like gain control and adaptation. However, recent studies showed that such processes fail to explain the pattern of color induction observed in homogeneous center-surround stimuli. Instead, color induction was suggested to result from scission: a contrast-dependent decomposition of the local image into two simultaneously present color components. Here, we show that scission can also induce large changes in the color appearance of randomly-variegated displays. We constructed center-surround displays in which the central target contained a distribution of random colors with an achromatic mean. The surround also contained a random distribution of colors, but the mean of this distribution was shifted in some direction away from the neutral point. These displays induce a clear percept of multiple layers: namely, a homogeneous colored target superimposed in a pattern of random noise. Asymmetric color-matching experiments revealed large opponent color induction effects in the center target region. The color-induction observed in our variegated displays far surpassed that observed in corresponding uniform center-surround stimuli. Importantly, strong color induction was only evident when the pattern of contrast along the border of the center pattern was compatible with scission, even if the same global color statistics were present. Color induction was also only observed if the chromatic contrast of the central region was less than that of the surround, consistent with models of transparency. These results cannot be accounted for by structure-blind adaptation or gain control mechanisms, and reveal that the importance scission in color induction is not restricted to homogeneous displays.
This PDF is available to Subscribers Only